Last week I published a screed about software politics. I received lots of comments — thank you all for that. The responses were, for the most part, thought-provoking and invigorating. There was one comment in particular which I feel compelled to address. Several people took me to task for saying a few words about the metonymy of “the press” in the expression “freedom of the press”, but failing to explore the rhetorical implications of the phrase “free speech”, thereby failing to address (in a rhetorical context) the oft-used construction “free as in speech”. They felt this considerably diluted my argument. I took this to mean that they were interested in a more robust exploration of the rhetorical and linguistic issues presented by the phrase “free as in speech, not free as in beer”.
First of all, let me make clear that this essay is not about software. This essay is about linguistics and rhetoric and epistemology. Here I am arguing that Richard Stallman may be like William of Ockham. Ockham is remembered today for his application and phrasing of the principle of parsimony — but no one remembers that he used what came to be known as his Razor in defense of the proposition that motion is a social construct; an argument for which he was tried for heresy (and acquitted). Following the trial, he stole the Seal of the Franciscans to sabotage the Pope’s authority and was excommunicated for that. Six hundred years later, all the great controversies of his life are mostly forgotten, except by medieval scholars, and he is remembered mostly for the phrasing of Occam’s Razor. (Ockham Explained) Similarly, the rhetorical construction around “free as in speech, not free as in beer” is sheer genius. Six hundred years from now, I predict, Richard Stallman will be remembered for those nine words, even if nobody remembers to what they refer.
The phrase is an anaphora containing ellipses, a zeugma, and is probably a schesis onamaton, although, as an amateur rhetorician, I would appreciate correction from anyone more skilled in the study of rhetoric than I.
Before diving into the meat of the discussion, let me recommend some background reading. The book I’m recommending is Understanding Computers and Cognition by Terry Winograd.
This is the book that kindled my interest in philosophy and magnified my interest in linguistics. Terry Winograd was one of the leading AI researchers of his time, and had written the best natural language recognition program of the era. In this book, he argues that all of the researchers attacking the problem of natural language recognition are doomed to fail because they are starting from an incorrect premise. That premise is that the meaning of a message is contained in the text of message. By analyzing the text of the message, one would be able to infer the meaning of the message. The point of the book was: this is not true. The meaning of the message is encoded both in the mental context of the sender of the message, and the mental context of the receiver of the message. His definition of “communication” is: “the thing that happens when you realize that what you thought you meant is not what they thought they heard.” (I may have paraphrased.) The negotiation of that contextual mismatch is “communication”.
It is precisely an understanding of this fundamental truth that leads to Lewis Carroll’s mockery, via the character of Humpty Dumpty in Through the Looking Glass, of people who think they can dictate what words or phrases mean based on the context of the speaker — ignoring the context of the listener. Dictionaries are not the way words acquire their meaning. Dictionaries are a compendium of ways in which people understand a word when it is used. And because there are lots of people understanding the same word in different ways, a dictionary will contain lots of different definitions of the same word. The act of defining a word is mostly collecting observations about what other people think the word means, not what the author of the dictionary thinks words mean. The Oxford English Dictionary is the acme of this endeavor. In my (abridged) copy, the definition of the word “free” spans five pages — and that’s before we get to phrases (“free and easy”) and compound words (“free lance”) and hyphenated words (“free-hand”) containing “free” which run on for another five pages. That’s ten pages of discussion about what “free” might mean. So when someone blurts out “free as in speech” that leaves quite a lot of nuance on the table. Let’s unpack some of that, shall we? I apologize in advance to non-native English speakers (and possibly non-American English speakers), as this is a linguistic exploration, the nature of the language and culture is very much at issue.
In order to internalize Terry Winograd’s observation about language, I’m going to propose the following gedanken experiment (as Ernst Mach would say).
Imagine, if you will, that you wish to crusade against political correctness by securing for all the freedom to flip people the bird as they see fit. This freedom you propose to call “bird freedom”. You will campaign for “Free Bird”. You publish a manifesto about “Free Bird”. You explain to your colleagues that by “Free Bird” you mean the freedom to flip the bird as you see fit. You ask them to join you in a campaign to collect signatures on a petition for “Free Bird”. The question you then ask them is: did they think of Lynyrd Skynyrd? The thought experiment is: how many times do you need to explain what you mean by “Free Bird” before they stop thinking of Lynyrd Skynyrd. How come they thought of Lynyrd Skynyrd? You never mentioned them! That’s your listener’s linguistic context digging in.
So let’s look at some “free” expressions that would seem to be relevant to a discourse about politics. In the U.S. there is often talk about “freedom of speech” and “freedom of assembly” and “freedom of religion” and “freedom of the press”. Oops — notice that an article crept in on that last one: that is where one clearly sees that “press” is different than “speech” — and the obvious part of that difference is called out by the addition of the article — the metonymy I mention in my previous screed. Phrases of the form “freedom of noun” are used as synonyms for “freedom of verb” where the noun-in-question is a synonym for “act-of-the-verb-in-question”. So, “freedom of the-act-of-speaking” is “freedom to speak”. “Freedom of the-act-of-assembling” is “freedom to assemble”. “Freedom of the-act-of-studying” is “freedom to study”. Simple enough. “Freedom of religion” is a little trickier. How do we transform that into “freedom to verb-meaning-religioning”? The usual phrase is “freedom to worship” or “freedom to practice religion”. Still works. One might try “freedom of belief” — but most people go with worship, because “freedom to believe” could refer to all kinds of belief, but “freedom to worship” zeroes in on religion. Also “freedom to believe” doesn’t necessarily imply that you could perform rituals, whereas “freedom to worship” does. “Freedom to practice (religion)” is another common phrasing.
“Freedom of ‘the press’” is trickier. The unpacked meaning would seem to be “freedom of the-act-of-publishing-the-news”. The verb form is “freedom to publish (the news)” and one infers that the reason it was not “freedom of publishing” is that this freedom was only intended to apply to journalism. The need for the metonym was a rhetorical construction to signal that the freedom in question applied only to people involved in “publishing-the-news” as opposed to anybody involved in “publishing-things-that-were-not-the-news”. The framing “freedom of ‘the press’” means “freedom to publish” whilst explicitly excluding people who might be trying to publish, say, software. From “freedom of noun”, we get to “free noun” — where “free” is an adjective modifying the noun. If the noun is any old noun, then the meaning is usually “free” as in “acquired without the exchange of currency”. If the noun is a synonym for “act-of-verbing”, then the adjective “free” can be exchanged for the adverb “freely” and the noun exchanged for the verb into the phrase “the right to freely verb”. So “free assembly” becomes “the right to freely assemble” and “free speech” means “the right to freely speak”. Although usually, it goes in the direction of “freedom of …”. Free press, of course, with that metonym thrown in, becomes “the right to freely publish-the-news”.
However, we can’t get to “adjective freedom” unless we have an adjective (freedom is the noun). So “freedom to worship” can become “freedom of religion”, but it can’t become “religion freedom”. It has to become “religious freedom”. “Freedom to learn” can become “freedom of education” but it usually turns into “academic freedom” rather than “education freedom”. Freedom of speech is rarely referred to as “speech freedom”. Likewise we don’t usually talk about “assembly freedom”. “Press freedom” is a tricky one. Press is not an adjective. What gives?
It is possible to turn a noun into adjective under certain (many) conditions. One such condition is if the noun represents a material (like, say wood). It can then be used as combinatorial adjective meaning made-of-noun ( wood instrument, wood inlay, or wood match — but not wood spoon — that one remains wooden spoon. Funnily enough “wooden spoon” means “trophy for finishing last.” Also, a spoon made out of wood). Nouns can also be turned into adjectives when being acted upon. So we turn the noun “rock” into an adjective as “rock borer”. In most of these “turn a material into an adjective” transformations, the result is a hyphenated word, or a compound word, not a phrase. You can also usually tell whether your adjective is formed via “made-of” or “acted-upon”. So a “wood skewer” would probably be a skewer made of wood, whereas a “wood plane” would probably be a plane used to shave wood.
In my last screed, I also pointed out that “freedom” was something that only applied to people, not things. The context in which I pointed this out was the political context. Freedom can apply to things, if it is in the context of motion (that’s why I brought up Ockham and his theory of motion) and friction. Wheels can turn freely — they have freedom of motion. (Motion, of course is a noun of the form “act-of-moving”). So wheels have freedom to move. Pendulums have freedom of motion, they can swing freely (where swinging, of course, is how pendulums move — as opposed to wheels which turn). However the kind of freedom implied by freedom of motion isn’t quite the nuance that we imagine when talking about “software freedom”. The phrase “free software means ‘free as in motion’” fails to impress those who have heard it.
Chickens might be able to range freely, they might have freedom to range. In which case, they would be free-range chickens. We don’t usually say “freedom of ranging” for chickens, because the transition from the noun phrase “of ranging” to the verb “to range” doesn’t work for most people. In fact, when these linguistic transformations fail to convince for whatever reason, the phrase gets turned into a single word or hyphenated word in order to distinguish it and insist that the meaning is distinct from what you would have thought based on the ordinary rules. Since “freedom of range” doesn’t work “free range” is problematic, and it enters the language as “free-range”. See free-wheeling and free-swinging and free-styling. The people using these phrases (sans hyphen) realize that the people hearing the phrase aren’t linguistically convinced, and need to make the phrase a word. One has a much better shot of convincing people that your word has a particular meaning, than to convince people that a collection of words means something other than what you’d expect given the definitions of all the constituent words.
Back to “press freedom”. How come that works, when “religion freedom” doesn’t? That’s the metonymy at work. Freedom (in the political sense, not the motion sense) acts upon people. Religion isn’t people. Assembly isn’t people. They are actions or beliefs that are involve people — but they are not people. But “press” is a metonym for “journalists”. Freedom can act upon people. So “press freedom” turns press into an adjective by the “freedom-which-acts-upon-journalists” transformation — where “press” means “journalists”. You could also understand it to be the “freedom-which-is-made-of-journalism” where “press” means “journalism” and get the same result! That metonym is what makes it work. As no such metonym is available for “religion”, say, “religion freedom” doesn’t work. It has to be “religious freedom”.
Now, rhetoric is an ancient discipline, and much more scientifically rigorous than software development. Pretty much every possible error you can make has a name, and you are allowed to make it, because that’s how language evolves. So, using a word in a context where it is the wrong part of speech (like a noun where you need an adjective) is called an “anthimeria”. “Religion freedom” is an anthimeria. You can say it. You can tell people what you mean when you say it. But the tricky bit is getting the people who hear your figure of speech to understand it to mean what you intended it to mean. They are free to understand it based on their education and context. Free as a bird. I predict their response will be: “Actually, it’s religious freedom”
Left as an exercise for the reader is to consider the phrase “software freedom”. “Software” is not an adjective. It is not a metonym for people. Perform the standard transformation and “freedom of software” seems incongruous because it is unclear what the act-of-softwaring is. What is the rhetorical device which channels meaning in the context of the receiver of the phrase?
With all of that background as prologue, let’s apply these principles to our favorite political software metaphor: beer. Imagine, if you will, that you are living in the U.S. during the time of Prohibition — alcohol is illegal. You wish to proselytize for the freedoms of the beer lover. Prohibition makes manufacturing, selling, or transporting beer illegal. Firstly, you want to promote the “freedom to sell beer”. This would be the “freedom of selling beer”.
The next freedom you wish to promote is the freedom for people to brew beer. This, too, is forbidden by Prohibition. You want the “freedom to brew (beer)” and the “freedom of brewing (beer)”. This could be referred to in slogans as “free brewing”. You run the risk of people confusing these phrases with the freedom to brew coffee, say, or tea. You want to avoid the discussions with trolls who are going to say “we already have the freedom to brew coffee and tea. We already have free brewing”. Your best strategy is to coin “beer-brewing” as a verb. “Freedom to beer-brew”. “Freedom of beer-brewing.” “Free beer-brewing”. Beer is a noun, but “brewing-which-acts-upon-beer” lets you use it as an adjective, and hyphenating it makes it a word. You got it immediately, right? I didn’t need to explain it. And similarly for the freedom to transport beer. Freedom of beer-shipping. Freedom to ship beer.
Armed with this terminology, we can attempt to lobby Congress to regain our freedoms vis-a-vis beer. But then, we aspire to rhetorical immortality. We conclude that in order to be most effective, we need all three of these freedoms. Any two might be progress, but not ENOUGH progress. We resolve to link the three inextricably. Either they all succeed together, or they all fail together. One of these freedoms alone is not worth having. How do we frame the dialog now? We can’t call it “freedom of beer” or “freedom to beer” because we are trying to ascribe three different verbs (selling, shipping, brewing) to the anthimeria “beer-as-verb”. There is no act-of-beering. There is act-of-selling, act-of-shipping, and act-of-brewing.
We decide to coin the phrase “beer freedom”. It’s an anthimeria. We obviously don’t mean “freedom made out of beer”. And, of course, we are not securing rights for beer, we are securing rights for people — and there is no metonym which turns “beer” into people. We remain undeterred. We’ll explain to people what we mean when we utter the apparently grammatically incorrect phrase “beer freedom”. Beer is not an adjective, so to use it as an adjective admits of a new definition. We have three different acts that we are contemplating, and we refuse to be deterred by the linguistic reality that typical phrases relating to freedom admit of only one such verb. “Beer freedom” consists of the three freedoms: the freedom to sell (beer), the freedom to ship (beer) and the freedom to brew (beer). All freedoms that exist for the verbs, but not when the object of the verb is beer. And in the pursuit of our “beer freedoms” we shall refer to beer that has these freedoms, beer which can legally be brewed and shipped and sold (like Canadian beer — there is no Prohibition in Canada); we shall refer to such beer as “free beer”.
It is possible that calling beer that respects “beer freedom” — “free beer” would lead to confusion. People would think they didn’t have to pay for “free beer”. Even if in your pamphlets and rhetoric you consistently point out that one of the “beer freedoms” is the freedom to sell the beer.
The opportunity for rhetorical immortality presents. “When we speak of ‘free beer’, we are referring to freedom, not price. We mean ‘free as in speech’, not ‘free as in …’”.
An uninspired English speaker (influenced, no doubt, by Lynyrd Skynyrd) would try to make their anti-Prohibition point by insisting that ‘free beer’ means ‘free as A bird’. It is an act of rhetorical genius to make the point by insisting that ‘free beer’ means ‘free as IN bird’.